Projects/paper suggestions CSCI6508 (Fall 2003)

You should explore a specific issue related to neurocomputing (either technical applications or neuroscientific models). You can thereby concentrate either on a critical analysis of the literature, or explore some of your ideas (though it has to include some literature discussion),

Here are some suggestions for themes:

• Load capacity of attractor networks (both ANN and CANN; follow Treves & Rolls, and study information content in CANN)
• Effective dimensionality discovery (with CANNs)
• Kai (Suggestion from my son)
• Learning modulation (Michael Hasselmo). See Hasselmo, M.E., Bodelon, C. and Wyble, B.P. (2002) A proposed function for hippocampal theta rhythm: Separate phases of encoding and retrieval enhance reversal of prior learning. Neural Computation, 14(4): 793-817.
  Appendix showing transition from equation 12 to equation 13.
• Conditioning and Neuromodulation (see Peter Dayan home page, particularly Dayan, P & Yu, A (2001).
  ACh, uncertainty, and cortical inference, NIPS 2001, to appear. See also Yu & Dayan (2001). )
• Classification (Grahame) (including ROC, see for example http://citeseer.nj.nec.com/provost97analysis.html)
• other topics you might be interested in

You will have to present your topic in class (~40min) and write it up in a format that I will specify later (conference paper style)

 Talk to me about your plans!